AI agents for e-commerce, manufacturing and engineering services
We put AI agents on the step that limits your business. Chosen for your constraints, built to run in production.
We map how you earn and spend money, find the constraint, and decide whether an agent, a model, plain software or no change at all is the right fix. Then we choose the tools and build it.
Industries
Three industries. We know where each one leaks.
E-commerce
D2C brands and marketplace sellers, from roughly ₹10 Cr / $1M a year in sales.
Where it leaks
- Visitors who leave with an unanswered question
- Abandoned carts and failed payments
- Returns and “where is my order” tickets
- Stock-outs on one channel, overstock on another
Agents we deploy
- Product and sizing assistant on site and WhatsApp
- Cart and payment recovery
- Order-status and returns agent
- Demand and reorder planning across channels
Manufacturing
Make-to-order and make-to-stock plants, and the distributors around them.
Where it leaks
- Quotes that take days because drawings are read by hand
- Purchase orders retyped from PDFs and email into the ERP
- Supplier follow-up done on the phone
- Production planned on spreadsheets
Agents we deploy
- RFQ-to-quote agent that reads drawings and specs
- Order-entry agent into your ERP
- Procurement agent for RFQs and supplier chasing
- Demand and replenishment planning
Engineering services
EPC and MEP contractors, design consultancies, fabrication and inspection firms.
Where it leaks
- Senior engineers’ hours lost to tender paperwork
- Proposals rewritten from scratch each time
- Drawings, RFIs and submittals tracked in email
- Billable hours and milestones invoiced late
Agents we deploy
- Tender screening and bid/no-bid agent
- Proposal drafting from your past bids
- Document-control and RFI tracking agent
- Timesheet-to-invoice and milestone billing
Where agents pay off
Two places to look: how money comes in, and how it goes out.
Revenue · the funnel
- 01Lead agent. Answers every inbound enquiry within a minute, qualifies it against your criteria, books the meeting.
- 02Follow-up agent. Sequences that never lapse, written from your CRM context.
- 03Payment recovery. Failed checkouts and renewals retried and chased automatically.
Cost · the chain
- 01Document agent. Reads invoices, purchase orders and shipping documents; posts them to your ERP.
- 02Reconciliation agent. Runs nightly; only the exceptions reach a person.
- 03Support agent. Resolves order-status and returns questions with real system access.
We measure every step first. The agent goes where the loss is largest; everywhere else, we leave things alone.
How we decide
Your constraints choose the tools. Not our preferences.
We track what is working across models, agent frameworks and infrastructure, and we have nothing to resell. The first deliverable is a decision in writing.
The written decision
- Constraint
- Where the money is lost, measured
- Fix
- Agent, model, software, or process change
- Stack
- Models, frameworks, hosting, build vs buy
- Return
- Expected value and how we'll measure it
- 01
Judge the problem
Map the process, measure the loss at each step, name the constraint.
- 02
Weigh your constraints
Budget, data quality, the systems you already run, your team's skills, compliance.
- 03
Choose the tools
Which model, which agent framework, build or buy, where a person stays in the loop.
- 04
Build and hand over
Secure by default, monitored, and runnable by your engineers or by us.
Results
Measured where it matters: cost of sale, margin, working capital.
−18%
Cost per sale
Funnel optimisation with high-quality win-back campaigns and upsells, so more of each sale came from customers already won.
+2.2 pts
Net margin
Customer credit-quality and recovery models for a large distributor, in an industry where margins are razor thin.
≈1 month
Working capital released each year
Procurement optimisation that cut wasted inventory, freeing almost a month of working capital annually.
Engagements
Start with a conversation. Commit only after the decision.
Free
30 min · memo in 24hOne process, mapped with the founder. You keep the memo.
1–2 weeks
fixed fee · credited to the buildThe written decision: constraint, fix, stack, return.
$15–50k
4–8 weeks · fixedProduction system with acceptance criteria agreed up front.
$1k/mo
ongoingWe monitor, retrain and improve it, or hand it to your team.
Who you'll work with
You'll speak with the person who builds it.
Dhrumil Barot has spent more than eight years in healthcare, e-commerce and manufacturing supply chains, and exited an AI startup in 2019, before the AI boom. He has built and scaled products across the whole customer journey, from acquisition and experience to retention, and on both sides of the P&L: revenue growth and cost optimisation. First with software, now with AI agents.
- Experience
- 8+ years
- Sectors
- Healthcare · E-commerce · Manufacturing supply chains
- Exit
- AI startup, 2019
Philosophy
Agents should add capacity, not only remove cost.
The larger return from an agent is usually the volume the same team can now handle: more orders through the same warehouse, more leads answered in the same hour, a new market opened without a new department. We design for that, and measure it.
Schedule a free consultation
Bring one process that costs you money or caps your revenue.
Thirty minutes with the founder. You leave with a view on the constraint; a written assessment follows within a day.
Free consultation · 30 min
Choose a day
Loading available times…
Prefer the full calendar? Open it on Cal.com